KC Posted July 18, 2019 Share Posted July 18, 2019 Last feb I had a power outage as multiple disks where writing to the server. Since then I've had 6095 errors in parity. Now my question - what is the order in which parity is written? 1: Write data disks then calculate parity? 2: Simultaneous Parity Calc? 3: Calc Parity before data is written. I'm just wonder what I should trust first, data disks or parity. Thanks for any help you can give. Quote Link to comment
itimpi Posted July 18, 2019 Share Posted July 18, 2019 You should trust parity! This is because although Unraid can detect that there is a parity mismatch it cannot tell which disk is causing it so you do not really have a lot of choice. There is always going to be a potential problem when the scenario you describe happens in that data being written at the moment of power loss can be lost. If in doubt you should run file system checks after such an occurrence on each disk to ensure it’s contents are in a consistent state. such occurrences are also the reason why a UPS is a highly recommended component in any Unraid system so the system can close itself in a tidy manner to minimise the chance of issues arising. Quote Link to comment
KC Posted July 18, 2019 Author Share Posted July 18, 2019 itimpi - Thank you for the reply. I do have a UPS - but it's too small to spin up the server and 16 drives (even one controller at a time) I had to remove it temporarily - which has become 8 months :( I actually did what you suggested and trusted parity - the array is currently rebuilding with only 4 hours to go. Quote Link to comment
itimpi Posted July 18, 2019 Share Posted July 18, 2019 1 minute ago, KC said: itimpi - Thank you for the reply. I do have a UPS - but it's too small to spin up the server and 16 drives (even one controller at a time) I had to remove it temporarily - which has become 8 months I actually did what you suggested and trusted parity - the array is currently rebuilding with only 4 hours to go. The file systems that Unraid uses are of the ‘journaling’ type so should be able to recover to a consistent state if needed. You are never going to know exactly what was being written at the moment of power loss anyway so the key thing is that already safely written data is intact. Quote Link to comment
KC Posted July 19, 2019 Author Share Posted July 19, 2019 (edited) Only 2 hours to go, and it corrected 6095 errors. Now if could just see which files the errors where corrected on. However that's a pretty big ask for Unraid. Thanks again for the input. I appreciate it. *** Edit Kind of makes me glad I did a MD5 on all files on my array - might take a week to figure out which (if any) files are damaged ... but I'll find em. And a side note - I don't mind thrashing my disks. I have spares. Edited July 19, 2019 by KC Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.